
% -------------------------------------------------------------------- PREAMBLE ------------------------------------------------------------------- %

\documentclass[12pt]{article}


\usepackage[T1]{fontenc}
\usepackage{amsfonts, amsmath, amssymb}
\usepackage{authblk}
\usepackage{multirow}
\usepackage{epsfig}
\usepackage{subfigure}
\usepackage{subfloat}
\usepackage{graphicx}
\usepackage{lscape}
\usepackage{bbm}
\usepackage{bm}
\usepackage[toc,page]{appendix}
\usepackage{titling}


\usepackage{booktabs}
\usepackage{longtable}
\usepackage{verbatim, rotating, paralist}
\usepackage{enumerate}
\usepackage{natbib}
\usepackage{soul}
\usepackage{caption}
\usepackage{blindtext}
\usepackage{multibib}


\usepackage{pdfsync}
\usepackage{latexsym}
\usepackage{amsthm}
\usepackage{mathabx}
\usepackage{mathabx}
\usepackage{stmaryrd}
\usepackage{dsfont}
\usepackage{tikz}
\usetikzlibrary{arrows,shapes,backgrounds,positioning,patterns,decorations.pathreplacing}


\usepackage{parskip}
\usepackage{anysize, indentfirst, setspace}
\usepackage[right=1.75cm, left=1.75cm, top=2.25cm, bottom=2.25cm]{geometry} 
\usepackage{epigraph}
\usepackage[toc,page]{appendix}
\usepackage{pdflscape}
\usepackage{fancyhdr}


\renewcommand{\topfraction}{.85}
\renewcommand{\bottomfraction}{.7}
\renewcommand{\textfraction}{.15}
\renewcommand{\floatpagefraction}{.66}
\renewcommand{\dbltopfraction}{.66}
\renewcommand{\dblfloatpagefraction}{.66}






% ------------------------------------------------------------- BEGIN DOCUMENT -----------------------------------------------------------%
\begin{document}


\Large{Main Manuscript}
\vspace{1.5in}





%% TABLE:  Results for Voter Registration
%%----------------------------------------------------
\begin{table}[h!]\centering\small
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\caption{Difference-in-Differences Results for Voter Registration}\label{Results Voter Registration} \vspace*{-.15in}
\vspace*{.057in}
\begin{tabular}{@{\extracolsep{5pt}}l*{6}{c}}
\noalign{\smallskip}\hline \hline\noalign{\smallskip} 
			&\multicolumn{3}{c}{Black Voter Registration}&\multicolumn{3}{c}{White Voter Registration}\\
			\cline{2-4} \cline{5-7}\noalign{\smallskip} \noalign{\smallskip}               
			\input{Registration_1_paper.tex} \\
			\noalign{\vspace*{-.18in}}
			\hline \hline\noalign{\smallskip} 
\multicolumn{7}{l}{\scriptsize Standard errors robust to clustering at the county level presented in parentheses.}\\
\multicolumn{7}{l}{\scriptsize \sym{*} \(p<0.1\), \sym{**} \(p<0.05\), \sym{***} \(p<0.01\)}\\
\multicolumn{7}{p{5.6in}}{\scriptsize{\emph{Notes:}  The table presents the estimates of $\beta$ from equation 1.  The dependent variable in specifications 1-3 is black voter registration; in specifications 4-6 it is white voter registration.  Outcomes are measured as a percent of the relevant voting age population.  An observation is a county-year pair.  All specifications include county and year fixed effects.  Controls1 includes: ln(population), \% population black, and median household income.  Controls2 also includes median years of schooling. Three counties lack black registration data for 1966 and are excluded from the analysis (see Online Appendix A).  An additional county lacks white registration data.  Estimates of control variable parameters are presented in Online Appendix E. }}
\end{tabular}
\end{table}




%% TABLE:  Results Voter Turnout
%%------------------------------------------
\begin{table}[h!]\centering\small
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\caption{Difference-in-Differences Results for Voter Turnout}\label{Results Voter Turnout} \vspace*{-.15in}
\vspace*{.057in}
\begin{tabular}{@{\extracolsep{5pt}}l*{6}{c}}
\noalign{\smallskip}\hline \hline\noalign{\smallskip} 
			&\multicolumn{3}{c}{Presidential Elections}&\multicolumn{3}{c}{Congressional Elections} \\
			\cline{2-4} \cline{5-7}\noalign{\smallskip} \noalign{\smallskip} 
                \input{Turnout_1_paper.tex} \\
                \noalign{\vspace*{-.18in}}
		\hline \hline\noalign{\smallskip} 
\multicolumn{7}{l}{\scriptsize Standard errors robust to clustering at the county level presented in parentheses.}\\
\multicolumn{7}{l}{\scriptsize \sym{*} \(p<0.1\), \sym{**} \(p<0.05\), \sym{***} \(p<0.01\)}\\
\multicolumn{7}{p{5.9in}}{\scriptsize{\emph{Notes:} The table presents the estimates of $\beta$ from equation 1.  The dependent variable in specifications 1-3 is voter turnout in presidential elections; in specifications 4-6 it is voter turnout in congressional elections.  Outcomes are measured as a percent of the relevant voting age population.  Voter turnout is not available broken out by race.  An observation is a county-year pair.  All specifications include county and year fixed effects. Controls1 includes: ln(population), \% population black, and median household income.  Controls2 also includes median years of schooling.  Estimates of control variable parameters are presented in Online Appendix E. }}
\end{tabular}
\end{table}



%% TABLE:  Results Democratic Vote Share
%%------------------------------------------------------
\begin{table}[h!]\centering\small
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\caption{Difference-in-Differences Results for Democratic Vote Share}\label{Results Democratic Vote Share} \vspace*{-.1in}
\vspace*{.057in}
\begin{tabular}{@{\extracolsep{5pt}}l*{6}{c}}
\noalign{\smallskip}\hline \hline\noalign{\smallskip} 
			&\multicolumn{3}{c}{Presidential Vote Share} &\multicolumn{3}{c}{Gubernatorial Vote Share}\\
			\cline{2-4}\cline{5-7}\noalign{\smallskip} \noalign{\smallskip} 
                  \input{Democrat_1_paper.tex} \\
                	\noalign{\vspace*{-.18in}}
		\hline \hline\noalign{\smallskip} 
\multicolumn{7}{p{6.2in}}{\scriptsize Standard errors robust to clustering at the county level presented in parentheses.}\\
\multicolumn{7}{l}{\scriptsize \sym{*} \(p<0.1\), \sym{**} \(p<0.05\), \sym{***} \(p<0.01\)}\\
\multicolumn{7}{p{6.2in}}{\scriptsize{\emph{Notes:}  The table presents the estimates of $\beta$ from equation 1.  The dependent variable in specifications 1-3 is presidential Democratic vote share, and in specifications 4-6 gubernatorial Democratic vote share.  An observation is a county-year pair.  All specifications include county and year fixed effects.  Controls1 includes: ln(population), \% population black, and median household income.  Controls2 also includes median years of schooling.  Estimates of control variable parameters are presented in Online Appendix E.  }}
\vspace{-.1in}
\end{tabular}
\end{table}



% ------------------------------------------------------------ ONLINE APPENDIX -------------------------------------------------------------- %
\clearpage
\newpage
\normalsize

\Large{Online Appendix}
\vspace{1.5in}


%% TABLE: Summary statistics
%%-------------------------------------
\begin{table}[h!]\centering\footnotesize \caption*{Table A1: Summary Statistics \label{sumstat}}
	\begin{tabular}{l c c c c c}\hline\hline \noalign{\smallskip}
	\multicolumn{1}{c}{Variable} & Mean
 		& SD & Min &  Max & N \\ 
		\hline \noalign{\smallskip}
	Black voter registration 				& \input{p_vreg_bnw_mean.tex} & \input{p_vreg_bnw_sd.tex} & \input{p_vreg_bnw_min.tex} & \input{p_vreg_bnw_max.tex} & \input{p_vreg_bnw_obs.tex}\\
	White voter registration    				& \input{p_vreg_w_mean.tex} & \input{p_vreg_w_sd.tex} & \input{p_vreg_w_min.tex} & \input{p_vreg_w_max.tex} & \input{p_vreg_w_obs.tex}\\
	White-black registration gap 			& \input{gap_bw_vreg_mean.tex} & \input{gap_bw_vreg_sd.tex} & \input{gap_bw_vreg_min.tex} & \input{gap_bw_vreg_max.tex} & \input{gap_bw_vreg_obs.tex}\\
	Voter turnout, presidential 				& \input{v_turnout_p_mean.tex} & \input{v_turnout_p_sd.tex} & \input{v_turnout_p_min.tex} & \input{v_turnout_p_max.tex} & \input{v_turnout_p_obs.tex}\\
	Voter turnout, congressional 			& \input{v_turnout_c_mean.tex} & \input{v_turnout_c_sd.tex} & \input{v_turnout_c_min.tex} & \input{v_turnout_c_max.tex} & \input{v_turnout_c_obs.tex}\\
	Democratic vote share, presidential 		& \input{vshare_p_dem_mean.tex} & \input{vshare_p_dem_sd.tex} & \input{vshare_p_dem_min.tex} & \input{vshare_p_dem_max.tex} & \input{vshare_p_dem_obs.tex}\\
	Democratic vote share, governor 		& \input{vshare_p_dem_gov_mean.tex} & \input{vshare_p_dem_gov_sd.tex} & \input{vshare_p_dem_gov_min.tex} & \input{vshare_p_dem_gov_max.tex} & \input{vshare_p_dem_gov_obs.tex}\\
	ln(population) 						& \input{pop_t_ln_mean.tex} & \input{pop_t_ln_sd.tex} & \input{pop_t_ln_min.tex} & \input{pop_t_ln_max.tex} & \input{pop_t_ln_obs.tex}\\
	\% population black 					& \input{pop_b_perc_mean.tex} & \input{pop_b_perc_sd.tex} & \input{pop_b_perc_min.tex} & \input{pop_b_perc_max.tex} & \input{pop_b_perc_obs.tex}\\
	Median household income 			& \input{inc_hh_med_mean.tex} & \input{inc_hh_med_sd.tex} & \input{inc_hh_med_min.tex} & \input{inc_hh_med_max.tex} & \input{inc_hh_med_obs.tex}\\
	Median years of schooling 			& \input{educ_25p_mean.tex} & \input{educ_25p_sd.tex} & \input{educ_25p_min.tex} & \input{educ_25p_max.tex} & \input{educ_25p_obs.tex}\\
\hline \hline \noalign{\smallskip}
\multicolumn{6}{p{4.6in}}{\scriptsize{\emph{Notes:}  Summary statistics in The table are presented for all counties in all years for which data for the variable is available.  Temporal data availability differs by measure.  See Table A2 and A3.  I present un-interpolated summary statistics for variables that are interpolated in the paper.  I exclude variables --- lynchings and incarceration rates --- that are not included in any of the main specifications of the table and are only measured in the pre-VRA period.  Table A4 includes these variables as it examines only pre-1965 levels.  An observation (N) is a county-year pair. The text describes the missing voter registration data.  There are also missing years of black population data as a result of suppression of small minority populations from the census data.  This is for one year only --- 1970 --- and I am able to interpolate for these missing observations using 1960 and 1980 data. }}
\end{tabular}
\end{table}





%% TABLE: Summary statistics by coverage status
%%----------------------------------------------------------------
\vspace{.1in}
\begin{table}[t!]\centering\footnotesize \caption*{Table A4: Summary Statistics by Coverage Status, Pre-1965 Levels \label{sumstat_coveragestatus}}
	\begin{tabular}{@{\extracolsep{5pt}} l c c c c    c c c c   }\hline\hline \noalign{\smallskip}\noalign{\smallskip}
				& \multicolumn{4}{c}{Covered} & \multicolumn{4}{c}{Uncovered} \\
				 \cline{2-5}  \cline{6-9} \\
	\multicolumn{1}{c}{Variable} & Mean & SD & Min &  Max &  Mean & SD & Min &  Max \\ 
		\hline \noalign{\smallskip}
	Black voter reg. 					& \input{p_vreg_bnw_cov_mean.tex} & \input{p_vreg_bnw_cov_sd.tex} &\input{p_vreg_bnw_cov_min.tex} & \input{p_vreg_bnw_cov_max.tex} & \input{p_vreg_bnw_uncov_mean.tex} & \input{p_vreg_bnw_uncov_sd.tex} &\input{p_vreg_bnw_uncov_min.tex} & \input{p_vreg_bnw_uncov_max.tex} \\
	
	White voter reg.    				& \input{p_vreg_w_cov_mean.tex} & \input{p_vreg_w_cov_sd.tex} &\input{p_vreg_w_cov_min.tex} & \input{p_vreg_w_cov_max.tex} & \input{p_vreg_w_uncov_mean.tex} & \input{p_vreg_w_uncov_sd.tex} &\input{p_vreg_w_uncov_min.tex} & \input{p_vreg_w_uncov_max.tex} \\
	
	White-black reg. gap 			& \input{gap_bw_vreg_cov_mean.tex} & \input{gap_bw_vreg_cov_sd.tex} &\input{gap_bw_vreg_cov_min.tex} & \input{gap_bw_vreg_cov_max.tex}& \input{gap_bw_vreg_uncov_mean.tex} & \input{gap_bw_vreg_uncov_sd.tex} &\input{gap_bw_vreg_uncov_min.tex} & \input{gap_bw_vreg_uncov_max.tex} \\
	
	Voter turnout, pres. 				& \input{v_turnout_p_cov_mean.tex} & \input{v_turnout_p_cov_sd.tex} &\input{v_turnout_p_cov_min.tex} & \input{v_turnout_p_cov_max.tex} & \input{v_turnout_p_uncov_mean.tex} & \input{v_turnout_p_uncov_sd.tex} &\input{v_turnout_p_uncov_min.tex} & \input{v_turnout_p_uncov_max.tex} \\
	
	Voter turnout, cong.				& \input{v_turnout_c_cov_mean.tex} & \input{v_turnout_c_cov_sd.tex} &\input{v_turnout_c_cov_min.tex} & \input{v_turnout_c_cov_max.tex} & \input{v_turnout_c_uncov_mean.tex} & \input{v_turnout_c_uncov_sd.tex} &\input{v_turnout_c_uncov_min.tex} & \input{v_turnout_c_uncov_max.tex} \\
	
	Dem. vote share, pres. 			& \input{vshare_p_dem_cov_mean.tex} & \input{vshare_p_dem_cov_sd.tex} &\input{vshare_p_dem_cov_min.tex} & \input{vshare_p_dem_cov_max.tex}  & \input{vshare_p_dem_uncov_mean.tex} & \input{vshare_p_dem_uncov_sd.tex} &\input{vshare_p_dem_uncov_min.tex} & \input{vshare_p_dem_uncov_max.tex}  \\
	
	Dem. vote share, gov. 			& \input{vshare_p_dem_gov_cov_mean.tex} & \input{vshare_p_dem_gov_cov_sd.tex} &\input{vshare_p_dem_gov_cov_min.tex} & \input{vshare_p_dem_gov_cov_max.tex}  & \input{vshare_p_dem_gov_uncov_mean.tex} & \input{vshare_p_dem_gov_uncov_sd.tex} &\input{vshare_p_dem_gov_uncov_min.tex} & \input{vshare_p_dem_gov_uncov_max.tex}  \\
	
	ln(population) 					& \input{pop_t_ln_cov_mean.tex} & \input{pop_t_ln_cov_sd.tex} &\input{pop_t_ln_cov_min.tex} & \input{pop_t_ln_cov_max.tex} & \input{pop_t_ln_uncov_mean.tex} & \input{pop_t_ln_uncov_sd.tex} &\input{pop_t_ln_uncov_min.tex} & \input{pop_t_ln_uncov_max.tex} \\
	
	\% population black 				& \input{pop_b_perc_cov_mean.tex} & \input{pop_b_perc_cov_sd.tex} &\input{pop_b_perc_cov_min.tex} & \input{pop_b_perc_cov_max.tex}& \input{pop_b_perc_uncov_mean.tex} & \input{pop_b_perc_uncov_sd.tex} &\input{pop_b_perc_uncov_min.tex} & \input{pop_b_perc_uncov_max.tex}  \\
	
	Median hh income 			& \input{inc_hh_med_cov_mean.tex} & \input{inc_hh_med_cov_sd.tex} &\input{inc_hh_med_cov_min.tex} & \input{inc_hh_med_cov_max.tex} & \input{inc_hh_med_uncov_mean.tex} & \input{inc_hh_med_uncov_sd.tex} &\input{inc_hh_med_uncov_min.tex} & \input{inc_hh_med_uncov_max.tex}  \\
	
	Median yrs schooling 			& \input{educ_25p_cov_mean.tex} & \input{educ_25p_cov_sd.tex} &\input{educ_25p_cov_min.tex} & \input{educ_25p_cov_max.tex} & \input{educ_25p_uncov_mean.tex} & \input{educ_25p_uncov_sd.tex} &\input{educ_25p_uncov_min.tex} & \input{educ_25p_uncov_max.tex}  \\

	
	Lynching victims  & \input{tot_victim_cov_mean.tex} & \input{tot_victim_cov_sd.tex} &\input{tot_victim_cov_min.tex} & \input{tot_victim_cov_max.tex} & \input{tot_victim_uncov_mean.tex} & \input{tot_victim_uncov_sd.tex} &\input{tot_victim_uncov_min.tex} & \input{tot_victim_uncov_max.tex} \\ 
	
	Black incarceration rate  & \input{intake_b_perc_cov_mean.tex} & \input{intake_b_perc_cov_sd.tex} &\input{intake_b_perc_cov_min.tex} & \input{intake_b_perc_cov_max.tex} & \input{intake_b_perc_uncov_mean.tex} & \input{intake_b_perc_uncov_sd.tex} &\input{intake_b_perc_uncov_min.tex} & \input{intake_b_perc_uncov_max.tex}  \\
	\noalign{\smallskip}
	Counties  & \multicolumn{4}{c}{40} & \multicolumn{4}{c}{60} \\
\hline \hline \noalign{\smallskip}
\multicolumn{9}{p{6.4in}}{\scriptsize{\emph{Notes:}  Summary statistics in the above table are presented for all counties in pre-1965 years for which data for the variable is available.  The number of years varies by measure.  See Table A3 for more details.   Note that these level differences in pre-VRA outcomes are accounted for in the estimation procedure by including county fixed effects.}}
\end{tabular}
\end{table}



%% TABLE:  Parallel Trends
%%---------------------------------
\vspace{.2in}
\begin{table}[h!]\centering\footnotesize 
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\caption*{Table C1: Results for Parallel Pre-Coverage Trends in Political Outcomes}\label{Parallel Trends Table}
\vspace*{.057in}
\begin{tabular}{l*{7}{c}}
\noalign{\smallskip}\hline \hline\noalign{\smallskip} 
 		&\multicolumn{1}{c}{Black}&\multicolumn{1}{c}{White}&\multicolumn{1}{c}{W-B}&\multicolumn{1}{c}{Pres.}&\multicolumn{1}{c}{Cong.}&\multicolumn{1}{c}{Pres. Dem.}&\multicolumn{1}{c}{Gov. Dem.}\\
                &\multicolumn{1}{c}{Voter Reg.}&\multicolumn{1}{c}{Voter Reg.}&\multicolumn{1}{c}{Gap}&\multicolumn{1}{c}{Turnout}&\multicolumn{1}{c}{Turnout}&\multicolumn{1}{c}{Vote Share }&\multicolumn{1}{c}{Vote Share }\\
                \input{Reg_ParallelTrends.tex}\\
                \noalign{\vspace*{-.19in}}
\noalign{\smallskip}\hline \hline\noalign{\smallskip} 
\multicolumn{8}{l}{\scriptsize Standard errors robust to clustering at the county level presented in parentheses.}\\
\multicolumn{8}{l}{\scriptsize \sym{*} \(p<0.1\), \sym{**} \(p<0.05\), \sym{***} \(p<0.01\)}\\
\multicolumn{8}{p{6.3in}}{\scriptsize{\emph{Notes:}  The table above presents results from estimating equation~\ref{Parallel Trends} restricted to the sample of years prior to 1965.  The number of sample years differs by outcome.  All specifications include county fixed effects.  All specifications include controls2.  An observation is a county-year pair.  }}
\end{tabular}
\end{table}



%% FIGURE: Voter Registration LATE
%%----------------------------------------------
\begin{figure}[h!]
	\begin{center}
	\vspace*{.2in} 
	\caption*{Figure D1: Voter Registration by Bandwidth around the Threshold for Section 5 Coverage}\label{LATE Registration}\vspace*{-.04in}
		\bigskip		
		Black Voter Registration \hspace*{1.27in} White Voter Registration \\
		\includegraphics[width=3.39in]{LATE_Registration_black.pdf} 
		\includegraphics[width=3.39in]{LATE_Registration_white.pdf}  \\ \smallskip
		 Gap in White-Black Registration  \\ 
		\includegraphics[width=3.39in]{LATE_Gap.pdf} \\ \smallskip
	\end{center}
\scriptsize{\emph{Notes:}  The figure above presents plots of the estimated difference-in-differences coefficients and 95\% confidence intervals from estimating equation 1 from the main body of the paper (including county and year fixed effects and using controls1) for different sized bandwidths around the 50\% threshold for Section 5 coverage.  Recall that coverage was determined by voter turnout rates in the 1964 presidential election.  Counties above 50\% turnout were left un-covered, while those below 50\% were covered.   The plot on the left shows these results for black voter registration as a percent of the black voting age population.  The plot on the right shows these results for the gap between white and black voter registration rates, each as a percent of the respective voting age population.  In addition, each plot shows the number of counties that fall into a given bandwidth around the threshold for coverage.  The x-axis in each plot is the bandwidth on either side of coverage.  For example, the value 0.2 on the x-axis corresponds to the bandwidth from 30\% voter turnout in the 1964 presidential election to 70\% (50\% $\pm20\%$).  Seventy of North Carolina's 100 counties had 1964 voter turnout rates in this range.    }
\end{figure} 




%% FIGURE: Turnout and Vote Share LATE
%%------------------------------------------------------
\begin{figure}[t!]
	\begin{center}
	\caption*{Figure D2: Voter Turnout and Democratic Vote Share by Bandwidth around the Threshold for Section 5 Coverage}\label{LATE Turnout}\vspace*{-.04in}
		\bigskip		
		Presidential Elections \hspace*{1.75in}  Congressional Elections \\
		\includegraphics[width=3.39in]{LATE_Turnout_P.pdf} 
		\includegraphics[width=3.39in]{LATE_Turnout_C.pdf} \\ \smallskip
		Presidential Democratic Vote Share \hspace*{.85in} Gubernatorial Democratic Vote Share \\
		\includegraphics[width=3.39in]{LATE_Democrat.pdf} 
		\includegraphics[width=3.39in]{LATE_Democrat_2.pdf} \\
		\smallskip
	\end{center}
\scriptsize{\emph{Notes:}  The figure above presents plots of the estimated difference-in-differences coefficients and 95\% confidence intervals from estimating equation 1 from the main body of the paper (including county and year fixed effects and using controls1) for different sized bandwidths around the 50\% threshold for Section 5 coverage.  Recall that coverage was determined by voter turnout rates in the 1964 presidential election.  Counties above 50\% turnout were left un-covered, while those below 50\% were covered.  The plot above-left shows these results for voter turnout in presidential elections as a percent of the voting age population.  The plot above-right shows these results for voter turnout in congressional elections as a percent of the voting age population.  The bottom left plot shows these results for vote share of the Democratic party in presidential elections, and the right plot Democratic party vote share in gubernatorial elections.  In addition, each plot shows the number of counties that fall into a given bandwidth around the threshold for coverage.  The x-axis in each plot is the bandwidth on either side of coverage.  For example, the value 0.2 on the x-axis corresponds to the bandwidth from 30\% voter turnout in the 1964 presidential election to 70\% (50\% $\pm20\%$).  Seventy of North Carolina's 100 counties had 1964 voter turnout rates in this range.    }
\end{figure} 





%% TABLE:  Results for Voter Registration
%%----------------------------------------------------
\begin{table}[h!]\centering\footnotesize
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\caption*{Table E1: Difference-in-Differences Results for Voter Registration}\label{Results Voter Registration Appendix} \vspace*{-.1in}
\vspace*{.057in}
\begin{tabular}{@{\extracolsep{5pt}}l*{6}{c}}
\noalign{\smallskip}\hline \hline\noalign{\smallskip} 
			&\multicolumn{3}{c}{Black Voter Registration}&\multicolumn{3}{c}{White Voter Registration}\\
			\cline{2-4} \cline{5-7}\noalign{\smallskip} \noalign{\smallskip} 
                \input{Registration_2_fullresults.tex} \\
                \noalign{\vspace*{-.19in}}
		\noalign{\smallskip}\hline \hline\noalign{\smallskip} 
\multicolumn{7}{l}{\scriptsize Standard errors robust to clustering at the county level presented in parentheses.}\\
\multicolumn{7}{l}{\scriptsize \sym{*} \(p<0.1\), \sym{**} \(p<0.05\), \sym{***} \(p<0.01\)}\\
\multicolumn{7}{p{5.7in}}{\scriptsize{\emph{Notes:}  The table presents the estimates of $\beta$ from estimating equation 1 in the main paper.  The dependent variable in specifications 1-3 is black voter registration, while the dependent variable in specifications 4-6 is white voter registration.  Each of these outcomes is measured as a percent of the relevant voting age population.  All specifications include county and year fixed effects.  An observation is a county-year pair. See Online Appendix~\ref{appendix_data} for more information on measurement and the sample size.}}
\end{tabular}
\end{table}


%% TABLE:  Results Voter Turnout
%%------------------------------------------
\begin{table}[h!]\centering\footnotesize
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\caption{Table E2: Difference-in-Differences Results for Voter Turnout}\label{Results Voter Turnout Appendix} \vspace*{-.1in}
\vspace*{.057in}
\begin{tabular}{@{\extracolsep{5pt}}l*{6}{c}}
\noalign{\smallskip}\hline \hline\noalign{\smallskip} 
			&\multicolumn{3}{c}{Presidential Elections}&\multicolumn{3}{c}{Congressional Elections} \\
			\cline{2-4} \cline{5-7}\noalign{\smallskip} \noalign{\smallskip} 
                \input{Turnout_2_fullresults.tex} \\
                \noalign{\vspace*{-.19in}}
		\noalign{\smallskip}\hline \hline\noalign{\smallskip} 
\multicolumn{7}{l}{\scriptsize Standard errors robust to clustering at the county level presented in parentheses.}\\
\multicolumn{7}{l}{\scriptsize \sym{*} \(p<0.1\), \sym{**} \(p<0.05\), \sym{***} \(p<0.01\)}\\
\multicolumn{7}{p{5.7in}}{\scriptsize{\emph{Notes:}  The table presents the estimates of $\beta$ from estimating equation 1 in the main paper.  The dependent variable in specifications 1-3 is voter turnout in presidential elections, while the dependent variable in specifications 4-6 is voter turnout in congressional elections.  Each of these outcomes is measured as a percent of the relevant voting age population.  All specifications include county and year fixed effects.  An observation is a county-year pair.  }}
\end{tabular}
\end{table}






%% TABLE:  Results Democratic Vote Share
%%-------------------------------------------------------
\begin{table}[h!]\centering\footnotesize
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\caption*{Table E3: Difference-in-Differences Results for Democratic Vote Share}\label{Results Democratic Vote Share Appendix} \vspace*{-.1in}
\vspace*{.057in}
\begin{tabular}{@{\extracolsep{5pt}}l*{6}{c}}
\noalign{\smallskip}\hline \hline\noalign{\smallskip} 
			&\multicolumn{3}{c}{Presidential Vote Share} &\multicolumn{3}{c}{Gubernatorial Vote Share}\\
			\cline{2-4}\cline{5-7}\noalign{\smallskip} \noalign{\smallskip} 
                \input{Democrat_2_fullresults.tex} \\
                \noalign{\vspace*{-.19in}}
		\noalign{\smallskip}\hline \hline\noalign{\smallskip} 
\multicolumn{7}{p{5.6in}}{\scriptsize Standard errors robust to clustering at the county level presented in parentheses.}\\
\multicolumn{7}{l}{\scriptsize \sym{*} \(p<0.1\), \sym{**} \(p<0.05\), \sym{***} \(p<0.01\)}\\
\multicolumn{7}{p{5.6in}}{\scriptsize{\emph{Notes:}  The table presents the estimates of $\beta$ from estimating equation 1 in the main paper.   The dependent variable in specifications 1-3 is two-party Democratic vote share in presidential elections; in specifications 4-6 it is for gubernatorial elections.  All specifications include county and year fixed effects.  An observation is a county-year pair. }}
\end{tabular}
\end{table}





%% TABLE:  Results White-Black Voter Registration Gap
%%-----------------------------------------------------------------------
\begin{table}[h!]\centering\footnotesize
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\caption*{Table E4: Difference-in-Differences Results for White-Black Voter Registration Gap}\label{Results Gap Registration Appendix} \vspace*{-.1in}
\vspace*{.057in}
\begin{tabular}{@{\extracolsep{5pt}}l*{3}{c}}
\noalign{\smallskip}\hline \hline\noalign{\smallskip} 
			&\multicolumn{3}{c}{W-B Registration Gap} \\
			\cline{2-4}\noalign{\smallskip} \noalign{\smallskip} 
                \input{Gap_1_fullresults.tex} \\
                \noalign{\vspace*{-.19in}}
		\noalign{\smallskip}\hline \hline\noalign{\smallskip} 
\multicolumn{4}{p{3.4in}}{\scriptsize Standard errors robust to clustering at the county level presented in parentheses.}\\
\multicolumn{4}{l}{\scriptsize \sym{*} \(p<0.1\), \sym{**} \(p<0.05\), \sym{***} \(p<0.01\)}\\
\multicolumn{4}{p{3.4in}}{\scriptsize{\emph{Notes:}  The table presents the estimates of $\beta$ from estimating equation 1 in the main paper.  The outcome is measured as the difference (white minus black) in registration rates each a percent of the relevant voting age population.  All specifications include county and year fixed effects.  An observation is a county-year pair.}}
\end{tabular}
\end{table}


%% FIGURE:  Registration
%%-------------------------------
\begin{figure}[h!]
	\begin{center}
		\caption*{Figure F1: Trends in Registration Rates By Coverage Status} \small
		\small \emph{Black registration increased by more in covered counties post-1965 than un-covered counties.  The white-black registration gap declined, but not by statistically significantly more in covered counties.} \\
		\smallskip\smallskip
			Black Registration \hspace*{1.75in} White Registration  \\
			\smallskip
			\includegraphics[clip=true, trim = 0in 0in 0in 0in, width=3.3in]{DinD_1_RegBlack.pdf}
			\includegraphics[clip=true, trim = 0in 0in 0in 0in, width=3.3in]{DinD_2_RegWhite.pdf}\\ 
			\smallskip White-Black Registration Gap \\
			\includegraphics[clip=true, trim = 0in 0in 0in 0in, width=3.3in]{DinD_3_Gap.pdf}
		\label{figure2}
		\end{center}
	\smallskip
	\scriptsize{\emph{Notes:} The plot above left presents trends in black voter registration as a percent of the relevant voting age population.  The plot above right presents trends white voter registration rates, each as a percent of the relevant voting age population.  The gap between white and black registration rates is the bottom plot.  Trends are averages fit separately to North Carolina counties covered by the 1965 VRA (solid line), and those un-covered (dashed line). The vertical line at 1965 indicates the introduction of Section 5 coverage. }
\end{figure}




%% FIGURE: Turnout
%%-------------------------
\begin{figure}[h!]
	\begin{center}
	\caption*{Figure F2: Trends in Turnout and Democratic Vote Share by Coverage Status}\label{figure3}\vspace*{-.04in}
		\small \emph{Voter turnout declined substantially in uncovered counties, but remained steady in covered counties.  Democratic vote share declined by more in covered than un-covered counties.} \\	
		\small	
		\smallskip \smallskip
		Presidential Turnout \hspace*{1.5in} Congressional Turnout \\
		\includegraphics[width=3.3in]{DinD_4_Turnout_P.pdf}
		\includegraphics[width=3.3in]{DinD_5_Turnout_C.pdf}\\
		\smallskip Democratic Presidential Vote Share \hspace*{.75in} Democratic Gubernatorial Vote Share \\
		\includegraphics[width=3.3in]{DinD_6_DemVShare.pdf}
		\includegraphics[width=3.3in]{DinD_7_DemVShareGov.pdf} \\
	\end{center}
\scriptsize{\emph{Notes:}  The plot above left presents trends in voter turnout in presidential elections as a percent of the relevant voting age population.  The plot above left presents trends in voter turnout in congressional elections as a percent of the relevant voting age population.  The plot below left presents trends in Democratic party vote share, and below right in gubernatorial Democratic vote share.  Trends are averages fit separately to North Carolina counties covered by the 1965 VRA (solid line), and those un-covered (dashed line).  The vertical line at 1965 indicates the introduction of Section 5 coverage. }
\end{figure}



%% TABLE:  Results for Voter Registration with alternative denominator
%%------------------------------------------------------------------------------------------
\begin{table}[h!]\centering\footnotesize
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\caption*{Table H1: Difference-in-Differences Results for Voter Registration Using Interpolated Voting Age Population}\label{Results Voter Registration Appendix2} \vspace*{-.1in}
\vspace*{.057in}
\begin{tabular}{@{\extracolsep{5pt}}l*{6}{c}}
\noalign{\smallskip}\hline \hline\noalign{\smallskip} 
			&\multicolumn{3}{c}{Black Voter Registration}&\multicolumn{3}{c}{White Voter Registration}\\
			\cline{2-4} \cline{5-7}\noalign{\smallskip} \noalign{\smallskip} 
                \input{Registration_3_altdenom.tex} \\
                \noalign{\vspace*{-.19in}}
		\noalign{\smallskip}\hline \hline\noalign{\smallskip} 
\multicolumn{7}{l}{\scriptsize Standard errors robust to clustering at the county level presented in parentheses.}\\
\multicolumn{7}{l}{\scriptsize \sym{*} \(p<0.1\), \sym{**} \(p<0.05\), \sym{***} \(p<0.01\)}\\
\multicolumn{7}{p{5.75in}}{\scriptsize{\emph{Notes:}  The table presents the estimates of $\beta$ from estimating equation 1 in the main paper.  The dependent variable in specifications 1-3 is black voter registration, while the dependent variable in specifications 4-6 white voter registration.  The outcome variables are calculated using interpolated values of the relevant voting age population to construct the denominator rather than using the percentages calculated by the sources described in Appendix~\ref{appendix_data}.  All specifications include county and year fixed effects.  An observation is a county-year pair. }}
\end{tabular}
\end{table}




%% TABLE:  Results White Voter Registration with alternative denominator
%%----------------------------------------------------------------------------------------------
\begin{table}[h!]\centering\footnotesize
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\caption*{Table H2: Difference-in-Differences Results for White-Black Voter Registration Gap}\label{Results Gap Registration Appendix2} \vspace*{-.1in}
\vspace*{.057in}
\begin{tabular}{@{\extracolsep{5pt}}l*{3}{c}}
\noalign{\smallskip}\hline \hline\noalign{\smallskip} 

                \input{Gap_2_altdenom.tex} \\
                \noalign{\vspace*{-.19in}}
		\noalign{\smallskip}\hline \hline\noalign{\smallskip} 
\multicolumn{4}{p{3.6in}}{\scriptsize Standard errors robust to clustering at the county level presented in parentheses.}\\
\multicolumn{4}{l}{\scriptsize \sym{*} \(p<0.1\), \sym{**} \(p<0.05\), \sym{***} \(p<0.01\)}\\
\multicolumn{4}{p{3.51in}}{\scriptsize{\emph{Notes:}  The table presents the estimates of $\beta$ from estimating equation 1 in the main paper.  The dependent variable is the white-black voter registration gap.  The outcome variables are calculated using interpolated values of the relevant voting age population to construct the denominator rather than using the percentages calculated by the sources described in Appendix~\ref{appendix_data}.  All specifications include county and year fixed effects.  An observation is a county-year pair.  }}
\end{tabular}
\end{table}







%% FIGURE: Simulation
%%----------------------------
\vspace{.1in}
\begin{figure}[h!]
	\begin{center}
	\caption*{Figure I1: Simulation of Black Voter Registration Responses to Black Migration}\label{simulation1}\vspace*{-.04in}
		\small	
		\smallskip \smallskip
		Uniform Percentage \hspace*{1.5in} Proportional to Black Population 
		\includegraphics[width=3.3in]{Simulation_Black_1.pdf}
		\includegraphics[width=3.3in]{Simulation_Black_2.pdf} \\
		\smallskip
		Proportional to Black Registration \\
		\includegraphics[width=3.3in]{Simulation_Black_3.pdf}
	\end{center}
\scriptsize{\emph{Notes:}  The above plots present simulations that re-estimate equation 1 from the paper using controls1, county as well as year fixed effects.  Each point estimate is a re-estimate of equation 1 with 90 and 95\% confidence bounds.  The plots also indicate the average total migrants that I simulate moving (by year) for a given uniform arrival percentage.  The plot above assumes that migrants left uncovered counties in uniform proportion and arrived to covered counties in uniform proportion.  The plot above right assumes migrants left uncovered counties in proportion to their black population, but arrived in covered counties in uniform proportion.  Finally the lower plot assumes migrants left uncovered counties in proportion to their black registration and arrived in covered counties in uniform proportion.     }
\end{figure}







%% TABLE: Duration Results
%%----------------------------------
\vspace{.2in}
\begin{table}[h!]\centering\footnotesize 
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\caption*{Table J1: Short, Medium and Long-Run Effects of Section 5 on Political Outcomes}\label{table_duration}
\vspace*{-.057in}
\begin{tabular}{l*{7}{c}}
\noalign{\smallskip}\hline \hline\noalign{\smallskip} 
 		&\multicolumn{1}{c}{Black}&\multicolumn{1}{c}{White}&\multicolumn{1}{c}{W-B}&\multicolumn{1}{c}{Pres.}&\multicolumn{1}{c}{Cong.}&\multicolumn{1}{c}{Pres. Dem.}&\multicolumn{1}{c}{Gov. Dem.}\\
                &\multicolumn{1}{c}{Voter Reg.}&\multicolumn{1}{c}{Voter Reg.}&\multicolumn{1}{c}{Gap}&\multicolumn{1}{c}{Turnout}&\multicolumn{1}{c}{Turnout}&\multicolumn{1}{c}{Vote Share }&\multicolumn{1}{c}{Vote Share }\\
                \input{Duration.tex}\\
                \noalign{\vspace*{-.19in}}
\noalign{\smallskip}\hline \hline\noalign{\smallskip} 
\multicolumn{8}{l}{\scriptsize Standard errors robust to clustering at the county level presented in parentheses.}\\
\multicolumn{8}{l}{\scriptsize \sym{*} \(p<0.1\), \sym{**} \(p<0.05\), \sym{***} \(p<0.01\)}\\
\multicolumn{8}{p{6.6in}}{\scriptsize{\emph{Notes:}  The table above presents results from estimating equation~\ref{duration}.  The number of sample years differs by outcome.  All specifications include county and year fixed effects.  All specifications include controls1.  An observation is a county-year pair.  There are not congressional turnout estimates post-1982 in this dataset.  }}
\end{tabular}
\end{table}





% ------------------------------------------------------------- END DOCUMENT -----------------------------------------------------------%
\end{document}




